ER-tethered stress sensor CREBH regulates mitochondrial unfolded protein response to maintain energy homeostasis Authors: Hyunbae Kim, Qi Chen, Donghong Ju, Neeraja Purandare, Xuequn Chen, Lobelia Samavati, Li Li, Ren Zhang, Lawrence I. Grossman, Kezhong Zhang Published: 2024 ABSTRACT SignificanceThis study defined that the endoplasmic reticulum (ER)-tethered stress sensor CREBH functions as a major regulator of mitochondrial homeostasis and energy metabolism by activating the mitochondrial Unfolded Protein Response (UPRmt) in response to energy demands. We identified a regulatory pathway originated at the ER-mitochondrial contact site, CREBH-ATF4/5-UPRmt, which modulates mitochondrial oxidative phosphorylation (OXPHOS) and glycolysis and thus rebalances the metabolic landscape and energy production under energy-demanding challenges. This work not only identified a master regulator of UPRmt, CREBH, but also revealed a crucial link between ER and mitochondria in maintaining mitochondria homeostasis and function under physiological stress conditions. CREBH (cAMP-responsive element-binding protein, hepatic-specific) is an endoplasmic reticulum (ER)-tethered stress sensor that plays a major role in maintaining energy homeostasis under stress conditions (1–4). In response to a variety of physiological ER stress signals such as inflammatory challenges, nutrient starvation, and circadian cues, CREBH transits from the ER to the Golgi where it goes through Regulated Intramembrane Proteolysis (RIP) to release a cleaved N-terminal fragment that functions as a basic leucine zipper (bZIP) transcription factor of the CREB/ATF family (5). The activated CREBH transcription factor drives expression of the genes encoding major metabolic regulators or enzymes involved in triglyceride (TG) lipolysis (1, 6), fatty acid oxidation (1, 7), lipogenesis (1, 2), gluconeogenesis (3, 8), glycogenolysis (3), and autophagy (9). Furthermore, the C-terminus of CREBH (CREBH-C), derived from the RIP, can be secreted into the extracellular space as a hepatokine upon energy demands or hepatic stress (4). CREBH-C regulates TG lipolysis and uptake by promoting lipoprotein lipase (LPL) activity through inhibitory interactions with angiopoietin-like (ANGPTL) 3 and ANGPTL8 in peripheral metabolic tissues. In humans, hypertriglyceridemia, atherosclerosis, or type-2 diabetes patients display high-rate nonsense mutations or rare genetic variants in the CREBH gene that encode CREBH N-terminal or C-terminal proteins (4, 6, 10). The Mitochondrial Unfolded Protein Response (UPRmt) is a mitochondria-originated pathway to maintain mitochondrial proteostasis during stress (11). Approximately 1,500 mitochondrial proteins encoded by mitochondrial or nuclear genomes need to be folded and assembled in the mitochondria under a high-fidelity control system to maintain proteostasis within the mitochondria (12). Upon accumulation of unfolded or misfolded proteins in the mitochondria, the UPRmt is activated to promote refolding or degradation of unfolded or misfolded proteins to maintain mitochondrial proteostasis (11). The UPRmt is a transcriptional activation program of mitochondrial chaperones and proteases initiated by mitochondria. Mitochondrial chaperone proteins involved in UPRmt mainly include heat shock protein (HSP) 10, HSP60, and Hsp70. Proteases, such as caseinolytic protease proteolytic (CLPP), mitochondrial Lon protease-like 1 (LonP1), and yme1-like 1 ATPase (YME1L1), can degrade proteins that are irreversibly damaged (13, 14). In mammals, UPRmt is regulated by bZIP transcription factors including ATF5 and ATF4 (15–17). As stress-inducible transcriptional regulators, ATF4 and ATF5 can act reciprocally or independently to regulate integrated stress response (ISR). Upon activation of the ATF5-mediated UPRmt, ATF4 regulates the transcriptional program of mitochondrial chaperone proteins and proteases by inducing the transcription of ATF5 (17, 18). However, under ER stress conditions, ATF4 can activate expression of the genes encoding functions involved in antioxidative-stress response or ER stress-associated apoptosis in a manner independent of ATF5 (19). The UPRmt establishes communication between the nucleus and mitochondria to facilitate mitochondrial proteomic surveillance and modulate expression of nuclear-encoded genes involved in reestablishing mitochondrial proteostasis (17). Mitochondrial Nuclear Retrograde Regulator 1 (MNRR1), a mitochondria-nuclear bio-organelle regulator, functions as a transcription factor to regulate UPRmt genes in the nucleus under stress conditions (20). It is increasingly recognized that UPRmt plays important regulatory roles in diverse physiological processes, particularly in metabolism and immunity (14). In response to disrupted mitochondrial proteostasis, the UPRmt can alter cellular energy metabolism by suppressing oxidative phosphorylation (OXPHOS) and increasing glycolysis (21, 22). Dysregulated UPRmt is associated with progression of metabolic disorders such as obesity, diabetes, and metabolic dysfunction-associated steatotic liver disease (MASLD) (14). In this study, we uncovered that the ER-tethered stress sensor CREBH functions as a transcription factor to regulate hepatic UPRmt in response to energy demands or under the control of the circadian clock. The CREBH-regulated UPRmt plays important roles in maintaining mitochondrial homeostasis and energy metabolism in the liver. Importantly, CREBH regulates ATF5 and ATF4, the master transcriptional regulators of the UPRmt, to promote a physiological UPRmt program independent of the ER-originated UPR (UPRER) pathway. CREBH deficiency leads to increased hepatic OXPHOS but deceased glycolysis. Our work depicts a molecular network where ER- and mitochondria-originated stress signaling pathways intertwine to regulate mitochondrial proteostasis and energy metabolism. Results CREBH Is Required to Maintain Mitochondrial Homeostasis and Function Upon Energy Demand. We previously identified that CREBH, an ER-resident latent transcription factor, is activated to regulate lipid and glucose metabolism in response to energy demands or hepatic stress (1–3, 7). To further understand the mechanistic basis by which CREBH regulates energy metabolism, we performed RNA-sequencing (RNA-seq) analysis with liver tissues of CREBH-KO and WT control mice under the energy demand triggered by fasting. The transcriptome analysis revealed remarkable hepatic genomic reprogramming in CREBH-KO mice after 14-h overnight fasting, in which 5 gene clusters were regulated by CREBH upon the fasting challenge (Fig. 1A). Comparison of gene expression profiles between CREBH-KO and WT mouse livers led to identification of 224 down-regulated and 187 up-regulated genes (with a cutoff of 2.0-fold change and FDR cutoff or adj P < 0.05) by CREBH abrogation under the fasting challenge (Fig. 1B). Gene Ontology (GO) pathway analysis revealed that CREBH regulates the genes encoding the components or functions of mitochondrial or ER membranes and mitochondrial matrix/homeostasis (Fig. 1C). Functionally, CREBH is involved in lipid and glucose metabolic pathways associated with mitochondrial functions (Fig. 1D). CREBH regulates expression of the genes encoding functions involved in mitochondrial homeostasis and energy metabolism. (A) K-means clustering heatmap analysis of genes differentially regulated in the livers of CREBH-KO mice after 14-h fasting. Each group is represented by RNA-seq data from three independent samples. Red, relative increase in abundance; green, relative decrease. Beside the heatmap is a T-SNE (Stochastic Neighbor Embedding) plot showing 5 gene clusters that are regulated by CREBH. N indicates the gene number for each cluster. (B) The Volcano Plot exhibiting the genes up- and down-regulated in the CREBH-KO mouse livers. RNA-seq analysis with liver RNAs isolated from CREBH-KO and WT mice after 14-h fasting identified 411 genes that were significantly down- or up-regulated (cutoff of 2.0-fold change and P < 0.05) in CREBH-KO mouse livers. (C and D) Gene ontology (GO) and pathway enrichment analyses of the genes that were significantly down-or up-regulated (cutoff of 2.0-fold change and P < 0.05) in CREBH-KO mouse livers. GO and pathway analyses were conducted using Enrichr and SRplot. (E–I) qPCR analyses of expression of the genes encoding functions involved in mitochondrial membrane and transport, mitochondrial chaperone machinery and proteases, mitochondria fatty acid (FA) metabolism, mitochondria fusion, and NAD kinase, as well as ER chaperones and ER stress markers in the livers of CREBH-KO and WT control mice under feeding or after 14-h fasting. The average mRNA level of the fed-WT mouse group was set as 1 and was used to calculate fold changes of mRNA levels in each group. Data are presented as mean ± SEM (n = 4). *P ≤ 0.05; **P ≤ 0.01. We performed quantitative real-time PCR (qPCR) analyses to verify expression of the genes encoding mitochondrial protein factors that play major roles in mitochondrial functions and homeostasis in the livers of CREBH-KO and WT control mice under the feeding condition or after fasting challenge. Consistent with the RNA-seq analysis results, expression levels of the following genes were significantly decreased in the CREBH-KO mouse livers under the fasting challenge (Fig. 1 E–I): 1) the genes encoding the functions involved in mitochondrial membrane components or protein transport, including TOMM7, TIMM23, Pam16, and ABCB10 (23); 2) the genes encoding mitochondrial protein chaperone machinery and proteases, including DNAJA3 (Hsp40), mt-Hsp70, Hspd1, and mt-HtrA2 (13); 3) the genes encoding the functions involved in mitochondrial fatty acid (FA) metabolism or beta oxidation, including Gcdh, Cpt1a, Slc27a1, Acsl1, and Acot2 (24); and 4) the genes encoding the major players in mitochondrial fusion and NAD kinase, including mitofusin (MFN) 1, MFN2, and mitochondrial NAD kinase (MNADK) (25, 26). In contrast, our qPCR analyses indicated that expression levels of the genes encoding the major ER chaperones or ER stress markers, including BiP/GRP78, ERdj4, P58ipk, EDEM1, and activated XBP1 (XBP1s), in CREBH-KO mouse livers were comparable to that in WT control mouse livers under the feeding condition or after the fasting challenge (Fig. 1I), implying that ER homeostasis or ER stress response in the liver was not affected by CREBH deficiency. Further, Western blot analyses confirmed that protein levels of the major enzymes or regulators involved in mitochondrial protein chaperone and folding machinery (Hsp10, DNAJA3, Hspd1, Hsp70), mitochondrial membranes and proteases (LonP1, CLPP), mitochondria FA oxidation (Cpt1α), as well as mitochondrial fusion and mitochondrial NAD kinase (MFN2, MNADK), were significantly decreased in CREBH-KO mouse livers under the fasting challenge (SI Appendix, Fig. S1 A and B). Together with the RNA-seq analysis, these results suggest that CREBH plays important regulatory roles in mitochondrial homeostasis and functions associated with mitochondrial membrane and proteostasis and in energy metabolism. Next, we examined the involvements of CREBH in mitochondrial homeostasis and functions by determining mitochondrial citrate synthase (CS) activities, mitochondrial membrane potential, ATP concentrations, cytochrome c oxidase (COX) activities, and reactive oxidative species (ROS) in the livers of CREBH-KO and WT control mice under the feeding condition or after 14-h fasting challenge. The activities of mitochondrial CS, an indication of inner mitochondrial membrane intactness and mitochondrial health (27), in the livers of CREBH-KO mice were significantly reduced compared to those of WT control mice regardless of the feeding or fasting condition (Fig. 2A). Consistently, the mitochondrial membrane potential in the livers of CREBH-KO mice was significantly decreased (Fig. 2B). Furthermore, the levels of mitochondrial ATP in CREBH-KO mouse livers were decreased, compared to that in WT control mouse livers where the fasting challenge increased mitochondrial ATP production (Fig. 2C). However, the activities of cytochrome c oxidase complex IV (COX) and complex I in the CREBH-KO mouse livers were significantly increased, compared to that in the control mouse livers, under the fasting challenge (Fig. 2 D and E), implicating an increase in mitochondrial respiratory electron transfer activity in the absence of CREBH. Further, coinciding with dysregulated mitochondrial activities, levels of total cellular ROS in the CREBH-KO mouse liver tissues were increased after the fasting challenge (Fig. 2 F and G), suggesting that CREBH deficiency may trigger increased ROS production from the cytosol while the mitochondrial membrane potential was decreased in the absence of CRBEH. Together, these measurements confirmed that CREBH is required to maintain mitochondrial homeostasis and function, especially under the fasting challenge. CREBH is required to maintain mitochondrial homeostasis and function upon energy demands. (A-E) Measurements of mitochondrial citrate synthase (CS) activities (A), mitochondrial membrane potential (aggregate vs monomer JC-1 ratios) (B), mitochondrial ATP concentrations (C), mitochondrial cytochrome c oxidase complex IV enzyme activities (D), and mitochondrial complex I activities (E) with intact mitochondria isolated from the livers of CREBH-KO and WT control mice under the feeding condition or after 14-h fasting using commercial enzymatic kits. Mitochondrial protein concentrations were determined for normalization of mitochondrial CS activities, ATP concentrations, as well as complex IV and complex I activities. For panels D and E, the average hepatic complex IV or complex I activity of the fed-WT mouse group was set as 100% and was used to calculate the percentages of complex IV or complex I activities for all the experimental groups. Data are presented as mean ± SEM (n = 4). ∗P ≤ 0.05; ∗∗P ≤ 0.01. (F) Levels of total cellular ROS in the liver tissues of CREBH-KO and WT control mice under the feeding condition or after 14-h fasting. Frozen liver tissue sections were stained with DHE Superoxide Indicator. Quantification of ROS signals, which were reflected by red fluorescence, in CREBH-KO and WT mouse liver tissues was performed for mean fluorescence intensity (MFI) and corrected total cell fluorescence (CTCF). The bar graph donates mean ± SEM (n = 4 mice). ∗∗P ≤ 0.01. (G) Representative cellular ROS staining in CREBH-KO and WT mouse liver tissues under the feeding or fasting condition as described in panel F. Scale bar, 20 µm. (H) TEM of liver tissue sections from CREBH-KO and WT control mice under feeding or after 14-h fasting. Scale bar, 500 nm. “E” indicates ER, “M” indicates mitochondria, and “LD” indicates lipid droplets. Beside the TEM images is the quantification of the number of MAM contact sites per cell (hepatocyte) and the percentage of MAM contacts vs total mitochondria number (MAM-forming mitochondria) per cell in the livers of WT or CREBH-KO mice under the feeding or fasting condition. Data presented as mean ± SEM (n = 6 for fed WT or KO group or 10 for fasted WT or KO group). ∗P ≤ 0.05; ∗∗P ≤ 0.01. CREBH Is Required for Fasting-Induced Expansion of Mitochondria-Associated Membranes (MAMs) in the Liver. To understand the molecular basis underlying dysregulated mitochondria homeostasis and function in CREBH-KO mouse livers, we assessed mitochondrial ultrastructure in the livers of CREBH-KO and WT control mice. Electron microscopy (EM) analysis revealed that CREBH deficiency led to disturbed mitochondrial ultrastructure, characterized by loss of cristae content, abnormal vacuolated structure, and electrondense aggregates in mouse liver hepatocytes (Fig. 2H). The morphological abnormality is consistent with the functional defects of CREBH-KO mitochondria, as reflected by the decreases in mitochondrial CS activities, membrane potential, and mitochondrial ATP production (Fig. 2 A–C). EM analyses also showed that the structure of ER-mitochondria contact sites, the ER-mitochondria intercompartment, namely mitochondrial-associated membranes (MAMs) (28), was diminished in CREBH-KO mouse livers, as indicated by the reduced number of MAM contact sites per cell and the percentage of MAM-forming mitochondria per cell in CREBH-KO mouse livers (Fig. 2H). To confirm this observation, we isolated MAM protein fractions from the livers of CREBH-KO and WT control mice under the feeding or fasting condition and examined stress-induced MAM components by Western blot analyses (Fig. 3 A–C and SI Appendix, Fig. S2). CREBH protein was detected in the purified MAM fraction of WT, but not KO, mouse livers (Fig. 3B). Importantly, the presence of CREBH in MAMs was remarkably increased in response to the fasting challenge. Further, the levels of mitofusin 2 (MFN2), a mitochondrial fusion protein responsible for tethering the ER and mitochondria during MAM formation (29), in MAMs of WT mouse livers were elevated in response to fasting, but not detectable in CREBH-KO MAMs regardless of the feeding or fasting condition (Fig. 3B). The chaperone glucose-regulated protein 75 (GRP75), a member of the heat shock protein 70 family expressed at the MAM interface, plays a key role in regulating ER-mitochondrial Ca2+ transfer (30). The level of GRP75 in the fasting-induced MAMs of CREBH-KO mouse livers was reduced, compared to that of WT mouse livers (Fig. 3B). However, the level of BiP/GRP78, an ER chaperone protein associated with MAMs, in the MAMs of CREBH-KO mouse livers was not significantly changed (Fig. 3B and SI Appendix, Fig. S2). CREBH regulates fasting-induced MAM components and prevents accumulation of unfolded proteins in mitochondria. (A) Illustration of fasting-induced hepatic MAMs and the approach to isolate purified MAM. (B and C) Western blot analyses of CREBH, MFN2, GRP75, GRP78, IRE1α, PERK, Sig-1R, and VDAC proteins in the purified MAM protein fractions isolated from the livers of CREBH-KO and WT mice under the feeding condition or after 14-h fasting. Five microgram (µg) purified MAM proteins per group was loaded for Western blot analysis. (D) Immunofluorescence labeling and confocal microscopy of the ER marker GRP78/BiP (green) and the MAM/mitochondrial marker MFN2 (red) in liver tissue sections from WT and CREBH-KO mice under the feeding or fasting condition. Scale bar, 100 μm; magnification: 400×. Images are representative of three independent experiments. (E) Quantification of ER-mitochondria membrane contact sites (MAMs) in the livers of CREBH-KO and WT mice under the feeding or fasting condition based on immunofluorescence labeling as described in the panel D. Quantification of GRP78-MFN2 merging fluorescent signals, which reflect MAM sites, in CREBH-KO and WT mouse liver tissues was performed for mean fluorescence intensity (MFI) and corrected total cell fluorescence (CTCF). The bar graph donates mean ± SEM (n = 4). ∗P ≤ 0.05; ∗∗P ≤ 0.01. (F and G) Quantitative analyses of TPE-MI-stained mitochondrial unfolded proteins in CREBH-KO and WT control livers. Mitochondria were isolated from the livers of CREBH-KO and WT mice under the feeding condition or after 14-h fasting, and incubated with TPE-MI dye, followed by measurement of fluorescence intensity for 2 h. Quantification of the maximal fluorescence ratios (panel G) was normalized to WT controls (1.0). Data are presented as mean ± SEM (n = 4 mice). ∗P ≤ 0.05; ∗∗P ≤ 0.01. (H) Representative TPE-MI staining of unfolded proteins in the mitochondria of WT and CREBH-KO mouse primary hepatocytes as well as WT primary hepatocytes overexpressing the activated CREBH (CREBH-A) or CRE control (CTL). Overexpression of CREBH-A or CRE control in mouse primary hepatocytes was achieved via adenoviral-mediated overexpression. WT or CREBH-KO primary hepatocytes were treated with 25 nM glucagon (Gluc) or PBS vehicle for 6 h, and then stained with MitoTracker (red) and TPE-MI (green) for imaging. Scale bar, 100 μm; magnification: 400×. We further examined the presence of IRE1α and PERK, the primary UPRER transducers that are involved in MAM formation under the challenge of cell survivability (31, 32), in the MAMs isolated from WT and CREBH-KO mouse livers under the feeding or fasting condition. The levels of IRE1α and PERK in the purified MAMs of WT mouse livers were decreased upon the fasting challenges (Fig. 3C and SI Appendix, Fig. S2). In comparison, the level of Sigma-1 Receptor (Sig-1R), a MAM-associated ER chaperone protein involved in calcium signaling through the IP3 receptor (33), in CREBH-KO mouse livers was not significantly changed, comparable to that in WT mouse livers (Fig. 3C). These data suggest that select MAM proteins, but not those involved in UPRER, are enriched in the hepatic MAMs in response to fasting. Supporting a requirement of CREBH for hepatic MAM enrichment, the levels of stress-induced MAM proteins, including MFN2, GRP75, IRE1α, and PERK, were markedly reduced in CREBH-KO mouse livers compared to that of the WT controls under the fasting challenge (Fig. 3 B and C and SI Appendix, Fig. S2). In contrast, the levels of the Voltage-Dependent Anion Channel (VDAC) protein were significantly increased in the MAMs isolated from CREBH-KO mouse livers (Fig. 3C), an interesting exception to be investigated in the future. We assessed formation of MAMs in the livers of WT and CREBH-KO mice by immunofluorescent costaining of the ER chaperone protein BiP/GRP78 and the mitochondrial fusion protein MFN2. Upon the fasting challenge, the MFN2 fluorescent signal was significantly elevated while the intensity of the BiP signal was barely changed in the WT mouse livers. Importantly, the merging fluorescent signals between BiP and MFN2, which reflect the formation of MAM contacts, were increased in WT, but not CREBH-KO, mouse livers in response to the fasting challenge (Fig. 3 D and E), thus confirming the requirement of CREBH for the expansion of fasting-induced hepatic MAMs. Taken together, these data suggest that CREBH preserves mitochondrial homeostasis in the liver and that CREBH is enriched in and required for hepatic MAM expansion in response to fasting. Loss of CREBH Elicits Accumulation of Unfolded Proteins in the Mitochondria. The involvement of CREBH in mitochondrial function and homeostasis as well as MAM enrichment upon fasting led us to speculate that CREBH may regulate mitochondrial proteostasis in response to an energy-demanding challenge. Indeed, our RNA-seq analysis showed that CREBH regulates expression of the genes encoding functions involved in mitochondrial proteostasis in mouse livers (Fig. 1). To provide direct support for this notion, we measured mitochondrial unfolded protein levels in CREBH-KO mouse livers. Tetraphenylethene maleimide (TPE-MI) is a turn-on fluorogenic dye used to probe unfolded proteins to report proteostasis imbalance (34). Mitochondria were isolated from the livers of WT and CREBH-KO mice under feeding or fasting condition, and then incubated with TPE-MI dye, followed by measurement of fluorescence intensity for 2 h. We observed an increased level of TPE-MI fluorescence, an indicator of higher unfolded protein abundance, in the mitochondria isolated from CREBH-KO mouse livers, compared to that in WT controls (Fig. 3 F and G). We assessed accumulation of mitochondrial proteins in CREBH-KO and WT mouse primary hepatocytes by confocal microscopy imaging for TPE-MI staining of the cells with mitochondrial tracker counterstain. Higher levels of unfolded proteins accumulated in the mitochondria of CREBH-KO primary hepatocytes, compared to that in WT primary hepatocytes, were observed (Fig. 3H). The accumulation of mitochondrial proteins in CREBH-KO primary hepatocytes, but not WT primary hepatocytes, was exacerbated upon the challenge of the fasting hormone, glucagon. Further, expression of the activated form of CRBEH (CREBH-A) relieved unfolded protein accumulation in the mitochondria (Fig. 3H). These results confirmed the role of CREBH in maintaining hepatic mitochondrial proteostasis under the energy demand challenge triggered by fasting or glucagon. CREBH Functions as a Transcriptional Regulator of UPRmt in the Liver in Response to Fasting or Under the Circadian Clock. To understand the mechanistic basis underlying the dysregulated mitochondrial homeostasis and function caused by CREBH deficiency, we examined whether CREBH regulates UPRmt, a major mitochondria-originated stress signaling pathway required to maintain mitochondrial proteostasis and function (35). We performed qPCR analyses to profile expression of the genes encoding mitochondrial protein factors that play major roles in UPRmt in the livers of CREBH-KO and WT mice under feeding or after fasting. Consistent with the RNA-seq analysis results (Fig. 1), expression levels of the following UPRmt hallmark genes were significantly decreased in the CREBH-KO mouse livers under the fasting challenge (Fig. 4 A–C): 1) the genes encoding the functions involved in mitochondrial proteases activity and mitochondrial proteome, including LonP1, CLPP, YME1L1, and PMPCB; 2) the genes encoding mitochondrial chaperones or protein folding enzymes, including Hsp10, Hsp60, and mt-Hsp70; 3) the genes encoding the functions involved in mitochondrial membrane components and mitochondrial protein transport, including TIMM17A and DNAJA3; and 4) the gene encoding NDUFB2 (NADH:Ubiquinone Oxidoreductase Subunit B2), a subunit of the mitochondrial respiratory chain Complex I. Further, Western blot analyses showed that protein levels of LonP1, CLPP, YME1L1, PMPCB, Hsp10, Hsp60, Hsp70, TIMM17A, DNAJA3, and NDUFB2 were significantly decreased in CREBH-KO mouse livers, compared to those in WT mouse livers, under the fasting challenge (Fig. 4D and SI Appendix, Fig. S3). CREBH regulates expression of the genes encoding major UPRmt components in the liver in response to fasting or under the circadian clock. (A–C) qPCR analyses of expression of the genes encoding major UPRmt components, including those involved in mitochondrial proteases and proteome (A), mitochondrial chaperones and folding enzymes (B), and mitochondrial membrane, transport, and respiration (C) in the livers of CREBH-KO and WT mice under feeding or after 14-h fasting. Data are presented as mean ± SEM (n = 4). *P ≤ 0.05; **P ≤ 0.01. (D) Western blot analyses of LonP1, CLPP, YME1L1, PMPCB, TIMM17A, Hsp10, Hsp60, Hsp70, DNAJA3, NDUFB2, and MTCO2 (internal control) proteins in purified mitochondrial protein fractions isolated from the livers of CREBH-KO and WT mice under the feeding or fasting condition. (E–N) Rhythmic expression levels of the genes encoding the major UPRmt components, including LonP1, CLPP, YME1L1, PMPCB, Hsp10, Hsp60, Hsp70, TIMM17A, DNAJA3, and NDUFB2, in the livers of CREBH KO and WT control mice across a 24-h circadian cycle. The mRNA expression levels were determined by qPCR. Fold changes in mRNA levels were determined by comparison to levels in one of the WT control mice at 6 PM. A, AM; P, PM. Data are presented as mean ± SEM (n = 3 mice per time point per genotype). *P ≤ 0.05; **P ≤ 0.01. Next, we examined whether CREBH regulates expression of the UPRmt hallmark genes in the liver under the circadian clock, a physiological oscillation cycle concomitant with rhythmic energy fluctuations. Expression of the genes encoding the mitochondrial proteases (LonP1, CLPP, YME1L1, PMPCB), mitochondrial protein chaperones (Hsp10, Hsp 60, mt-Hsp70), and mitochondrial membrane protein transporters or respiratory components (TIMM17A, DNAJA3, NDUFB2) in WT mouse livers exhibited circadian rhythmicity across a 24-h circadian cycle (Fig. 4 E–N). In contrast, circadian expression levels and rhythmicity of the major UPRmt genes in CREBH-KO mouse livers were decreased or disrupted, compared to those in WT mouse livers, across the circadian cycle (Fig. 4 E–N), suggesting that CREBH functions as a transcriptional regulator of the hepatic UPRmt program under the circadian clock. CREBH Activates Expression of Hepatic ATF5 and ATF4, the Master Transcriptional Regulators of UPRmt, in an UPRER-Independent Manner. We sought to determine the mechanistic basis by which CREBH regulates hepatic UPRmt in response to energy demands. The qPCR analyses showed that expression levels of the mRNAs encoding the major UPRmt transcriptional regulators ATF4 and ATF5, but not CHOP or ATF3, were significantly reduced in the livers of CREBH-KO mice, compared to those of WT mice (Fig. 5 A–D). However, fasting-induced expression of ATF4 and ATF5, but not ATF3, in mouse livers was diminished in the absence of CREBH (Fig. 5 A, B, and D). Consistent with the gene expression profile, Western blot analyses confirmed that the protein levels of ATF4 and ATF5 were decreased in the livers of CREBH-KO mice compared to these of WT control mice (Fig. 5 E and F and SI Appendix, Fig. S4). These results indicated that CREBH is required for the induction of hepatic ATF4 and ATF5 in response to fasting. CREBH activates expression of hepatic ATF5 and ATF4 in an UPRER-independent manner. (A–D) qPCR analyses of expression of the genes encoding ATF5, ATF4, CHOP, and ATF3 in the livers of CREBH-KO and WT mice under feeding or after 14-h fasting. The average mRNA level of the fed-WT mouse group was set as 1 and was used to calculate fold changes of mRNA levels in each group. Data are presented as mean ± SEM (n = 4). *P ≤ 0.05; **P ≤ 0.01. (E and F) Western blot analyses of ATF5, ATF4, ATF3, CHOP, and GAPDH proteins in the total protein lysates isolated from the livers of CREBH-KO and WT mice under the feeding or fasting condition. (G–I) Rhythmic expression levels of the mRNAs encoding ATF5, ATF4, and CHOP in the livers of CREBH KO and WT mice across a 24-h circadian cycle. The mRNA levels were determined by qPCR. Fold changes in mRNA levels were determined by comparison to the level in one of the WT control mice at 6 PM. A, AM; P, PM. Data are presented as mean ± SEM (n = 3 mice per time point per genotype). *P ≤ 0.05; **P ≤ 0.01. (J) Western blot analyses of ATF5, ATF4, and GAPDH proteins in the livers of CREBH-KO and WT mice across a 24-h circadian cycle. Each sample per genotype per time point was pooled from 3-5 mouse liver protein lysates. We next examined whether CREBH regulates rhythmic expression of ATF5 and ATF4 in mouse livers across the circadian clock. Rhythmic expression of the ATF5 and ATF4 mRNAs in CREBH-KO mouse livers was repressed, compared to that in WT control mice, across the 24-h circadian cycle (Fig. 5 G and H). As a control, circadian rhythmic expression profiles of CHOP in CREBH-KO and WT mouse livers are comparable (Fig. 5I). Further, Western blot analyses demonstrated that expression of ATF5 and ATF4 proteins in mouse livers was diminished in the absence of CREBH across the circadian clock (Fig. 5J and SI Appendix, Fig. S5A). It was reported that the PERK-ATF4-CHOP branch of UPRER promotes UPRmt signaling in response to pathologic insults that induce ER stress (36). We investigated whether CREBH interacts with the PERK-mediated UPRER signaling to regulate the physiological UPRmt triggered by fasting. Western blot analyses showed that the levels of phosphorylated PERK or eIF2α were not increased in the livers of fasted WT mice or CREBH-KO mice regardless of feeding or fasting (SI Appendix, Fig. S5 B and C). Further, the protein levels of CHOP and the activated form of XBP1 (XBP1s), two downstream regulators of UPRER signaling, were not altered in WT or CREBH-KO mouse livers under the feeding or fasting conditions (SI Appendix, Fig. S5 B and C), suggesting a lack of ER stress response or UPRER activation in the liver in response to the energy-demanding challenge or under CREBH deficiency. This is consistent with the observation that expression levels of the genes encoding the major ER chaperones or ER stress markers in CREBH-KO mouse livers were comparable to that of WT mouse livers under the feeding or fasting condition (Fig. 1I). Taken together, our studies demonstrated that CREBH regulates fasting-induced or circadian rhythmic expression of ATF5 and ATF4, the major UPRmt transcriptional regulators, in the liver in a manner independent of UPRER. CREBH Directly Binds to the ATF4 and ATF5 Gene Promoters in the Liver. To understand the molecular basis by which CREBH promotes fasting-induced expression of ATF5 and ATF4, we analyzed cis-regulatory elements in the promoter regions of the ATF5 and ATF4 genes. The mouse ATF5 or ATF4 gene promoter region possesses multiple transcriptional binding sites matched to the CREB binding motif (CRE) or PPARα binding motif (PPRE) (SI Appendix, Fig. S6). Particularly, there exists an integrated CRE-PPRE binding motif 4 bp away from the ATF5 gene transcription start site. We originally identified the integrated CRE-PPRE element present in the promoter region of the FGF21 gene, which is bound by a CREBH-PPARα transcriptional complex to activate maximal expression of FGF21 in response to fasting (7). To test whether CREBH and/or PPARα bind to and activate the ATF4 and ATF5 gene promoters, we performed chromatin immunoprecipitation (ChIP)-qPCR analysis to determine endogenous CREBH or PPARα enrichment in the promoter regions of the ATF5 and ATF4 genes in the livers of the mice under the feeding or fasting condition. ChIP-qPCR analyses indicated that both CREBH and PPARα bound to the ATF5 gene promoter containing the CRE-PPRE element in the WT mouse livers (Fig. 6A). In response to the fasting challenge, the binding activities of hepatic CREBH and PPARα to the ATF5 gene promoter were significantly increased (Fig. 6A). Further, the binding activity of CREBH to the ATF4 gene promoter containing the CRE sequence located at 7 bp away from the transcription start site was detected in the WT mouse livers, and this binding activity was significantly increased after the fasting challenge (Fig. 6B). However, no binding enrichment of PPARα to the ATF4 gene promoter regions containing the PPRE or CRE elements was detected in the WT mouse livers regardless of the feeding or fasting condition (Fig. 6B and SI Appendix, Fig. S6). CREBH and PPARα bind to the promoter and activate expression of ATF5 or ATF4 gene in the liver. (A and B) Binding activities of endogenous CREBH or PPARα to the ATF5 or ATF4 gene promoter region containing the CRE-PPRE or CRE element in WT or CREBH-KO mouse livers under feeding or after 14-h fasting determined by ChIP-qPCR. Fold changes of CREBH or PPARα binding enrichment in the gene promoters under the feeding or fasting condition were quantified by comparing ChIP-qPCR signals from the samples pulled down by the anti-CREBH or anti-PPARα antibody with that pulled down by the anti-IgG antibody. The average fold change of CREBH or PPARα binding enrichment in the CREBH-KO group was set as 1 and was used to calculate the fold changes of CREBH or PPARα binding activities for each group. Data are presented as mean ± SEM (n = 3 animals per group). *P ≤ 0.05. CKO, CREBH-KO; FE, feeding; FA, fasting; IP-CREBH/ PPARα, mouse liver chromatin pulled down by the anti-CREBH/anti-PPARα antibody. (C and D) ChIP analysis of CREBH and PPARα binding activities to the endogenous ATF5 or ATF4 gene promoter in the livers of CREBH-KO (CKO), PPARα-KO (PKO), and WT mice expressing Flag-tagged CREBH, Flag-tagged PPARα, or GFP control under feeding or after 14-h fasting. For ChIP-PCR, mouse chromatins were immunoprecipitated with the antibody against Flag. Nonimmunoprecipitated chromatins were included as input controls. The sequence information of the primers amplifying the CRE or PPRE-binding motifs in the ATF5 or ATF4 gene promoter is in SI Appendix, Table S2. The amplified PCR products were visualized on agarose gels. (E) qPCR analyses of expression of the genes encoding ATF5 and ATF4 in the livers of PPARα-KO and WT mice under feeding or after 14-h fasting. Data are presented as mean ± SEM (n = 3). *P ≤ 0.05. PKO, PPARα-KO. (F) Western blot analyses of ATF5, ATF4, and GAPDH proteins in the livers of PPARα-KO and WT mice under feeding or after 14-h fasting. To further delineate the binding activities of CREBH and PPARα to the ATF5 and ATF4 gene promoters, we performed ChIP assays with WT, CREBH-KO, and PPARα-KO mouse livers expressing Flag-tagged CREBH, Flag-tagged PPARα, or GFP. In the Flag antibody-pulled down chromatin of WT mouse livers expressing Flag-tagged CREBH, the ATF5 promoter region containing the CRE-PPRE, but not the other CRE motifs, can be amplified by a ChIP-PCR (Fig. 6C, lanes 3 and 4), indicating the direct binding activity of CREBH to the CRE-PPRE-containing ATF5 promoter. In response to fasting challenge, the binding activity of CREBH to the ATF5 gene promoter was increased (Fig. 6C, lane 4), compared to that under the feeding condition. In the Flag antibody-pulled down chromatin of WT or CREBH-KO mouse livers expressing Flag-tagged PPARα, the ATF5 promoter region containing the CRE-PPRE motifs can be amplified by ChIP-PCR (Fig. 6C, lanes 5 and 7), indicating the binding activity of PPARα to the CRE-PPRE site of the ATF5 gene promoter. Additionally, CREBH binding activity to the CRE-PPRE-containing ATF5 gene promoter region can be detected in the pull-down chromatin of PPARα-KO mouse livers expressing Flag-tagged CREBH (Fig. 6C, lane 6). ChIP-PCR assay was able to detect binding activities of CREBH to the ATF4 gene promoter region containing the CRE element close to the transcription start site in the WT or PPARα-KO mouse livers expressing CREBH (Fig. 6D, lanes 3, 4, and 6). Upon fasting, the binding activity of CREBH to the ATF4 gene promoter was increased in the mouse livers (Fig. 6D, lane 4), compared to that under the feeding condition. In contrast, no binding activity of PPARα to the ATF4 gene promoter can be detected with the WT or CREBH-KO mouse livers expressing PPARα (Fig. 6D, lanes 5 and 7), suggesting that PPARα does not function as a transcription factor to activate ATF4 expression. PPARα Regulates Expression of Hepatic ATF5, but not ATF4, in a CREBH-Dependent Manner. As PPARα bound to the ATF5, but not ATF4, gene promoter in mouse livers after fasting (Fig. 6 A–D), we determined whether PPARα regulates ATF5 expression in mouse livers. The qPCR analyses showed that fasting-induced expression of the ATF5 mRNA was significantly reduced in the livers of PPARα-KO mice, compared to those of WT mice (Fig. 6E). In contrast, expression levels of the ATF4 mRNA in the livers of WT and PPARα-KO mice were comparable under the feeding or fasting condition (Fig. 6E), suggesting that PPARα regulates expression of ATF5, but not ATF4, in the livers upon fasting. Consistently, Western blot analyses confirmed that the protein levels of ATF5, but not ATF4, were decreased in the livers of PPARα-KO mice compared to these of WT control mice (Fig. 6F and SI Appendix, Fig. S7 A and B). CREBH regulates and interacts with PPARα to drive expression of the fasting hormone FGF21 and the key autophagy genes in the liver upon nutrient starvation (7, 9). We tested the interdependent relationship between CREBH and PPARα in regulating ATF5 by examining expression of ATF5 in PPARα-KO mice expressing exogenous CREBH and in CREBH-KO mice expressing exogenous PPARα. Supporting the role of PPARα or CREBH in regulating ATF5 expression, levels of hepatic ATF5 mRNA were reduced in either PPARα-KO or CREBH-KO mice compared with those in the control mice under the feeding or fasting condition (SI Appendix, Fig. S7 C and D). Interestingly, expression of exogenous CREBH, but not GFP, increased expression of ATF5 in the PPARα-KO mice upon fasting (SI Appendix, Fig. S7C). In contrast, expression of exogenous PPARα failed to rescue reduced ATF5 expression in the CREBH-KO mice (SI Appendix, Fig. S7D), suggesting that CREBH plays a more determinant role in driving ATF5 expression and that PPARα likely relies on CREBH to exert its trans-activation effect on ATF5 in the liver upon fasting. Additionally, we examined expression of ATF4 in PPARα-KO mice expressing exogenous CREBH or in CREBH-KO mice expressing exogenous PPARα. Although expression of exogenous CREBH can sustain, even increase, expression of ATF4 in PPARα-KO mouse livers, PPARα failed to induce ATF4 expression in the CREBH-KO mice (SI Appendix, Fig. S7 E and F), confirming that PPARα has no regulatory role in hepatic ATF4 expression. CREBH Regulates Hepatic Mitochondrial Energy Metabolism via the UPRmt -OXPHOS-Glycolysis Regulatory Cascade. UPRmt is known to suppress the transcription of mitochondrial oxidative phosphorylation (OXPHOS) components and repress COX activities to reduce the load on the mitochondrial proteostasis machinery during stress (15, 22). Concomitantly, UPRmt induces glycolysis gene expression and increases glycolysis activities, thereby rebalancing the metabolic landscape of the cell by promoting an alternative means of ATP synthesis. As CREBH deficiency leads to defective hepatic UPRmt, reduced mitochondrial ATP production, and increased COX activities (Figs. 2 C–E and 4), we further evaluated whether CREBH deficiency affects OXPHOS, glycolysis, and mitochondrial energy metabolism in the liver. First, we analyzed expression of the nuclear genes encoding the cytochrome c oxidase (COX) subunits of mitochondrial respiratory chain complex IV, which plays a major regulatory role in OXPHOS, in CREBH-KO and WT mouse livers. Expression levels of the genes encoding the nuclear encoded COX components of complex IV, including COX4I1, COX5A, COX6A1, and COX7A1, were significantly increased in the livers of CREBH-KO mice, compared to those in WT control mice (Fig. 7A), suggesting an elevation of hepatic OXPHOS in the absence of CREBH. Next, we examined expression of the genes encoding major enzymes or regulators in glycolysis, including Aldob, LDHA, Pdhb, Glut1, and Pdk4, in the livers of CREBH-KO and WT mice. Expression levels of these glycolysis genes in CREBH-KO mouse livers were significantly decreased compared to those in the WT control mouse livers (Fig. 7B), indicating that CREBH deficiency led to decreased hepatic glycolysis. Notably, expression of the gene encoding pyruvate dehydrogenase kinase 4 (PDK4), a vital mitochondrial protein controlling the switch between glycolysis and OXPHOS (37), was suppressed in the CREBH-KO mouse liver regardless of feeding or fasting challenge (Fig. 7B). Further, increased hepatic OXPHOS in the absence of CREBH was confirmed by the elevation of COX4I1 and MTCO1 proteins, the subunits of mitochondrial complex IV and complex I, in the mitochondria fractions isolated from CREBH-KO mouse livers (Fig. 7C and SI Appendix, Fig. S8). CREBH regulates hepatic mitochondrial energy metabolism by modulating the UPRmt -OXPHOS-Glycolysis regulatory cascade. (A and B) qPCR analyses of expression levels of the genes encoding selected subunits of COX (A) and the genes encoding the major enzymes or regulators in glycolysis (B) in the livers of CREBH-KO and WT mice under feeding or after 14-h fasting. Data are presented as mean ± SEM. *P ≤ 0.05; **P ≤ 0.01. (C) Western blot analyses of representative components of COX, including COX4l1 (COX4), MTCO1, NDUFS3, and MTCO2 in the purified mitochondria isolated from the livers of PPARα-KO and WT mice under feeding or after 14-h fasting. The levels of MTCO2, a COX2 subunit, were determined as an internal control. (D–F) Mitochondrial bioenergetic profiling of CREBH-KO and WT mouse primary hepatocytes. The mitochondrial bioenergetics of hepatocytes were profiled by measuring oxygen consumption rate (OCR) using Seahorse XF Analyzer. A mitochondrial stress test was performed using a sequential injection of oligomycin (1.5 μM), FCCP (1 μM), and rotenone/Antimycin A (1 μM each). Basal mitochondrial OCR was normalized by subtracting nonmitochondrial OCR obtained after adding rotenone/antimycin. Maximal OCR was calculated by subtracting nonmitochondrial OCR from the highest OCR measurement after adding FCCP. R/A, rotenone/antimycin. Data are presented as mean ± SEM (n = 6). ∗∗P ≤ 0.01. (G) Rate of glycolysis of CREBH-KO and WT mouse primary hepatocytes determined by Seahorse analyzer. The Seahorse Analyzer provides kinetic readouts of OCR and extracellular acidification rate (ECAR), which are used to determine the glycolytic proton efflux rate (glycolysis) of the cells. The data are presented as mean ± SEM (n = 6). ∗∗P ≤ 0.01. (H) Illustration of the pathways by which CREBH regulates a physiological UPRmt as well as mitochondrial homeostasis and energy metabolism in response to energy demands. Next, we profiled mitochondrial bioenergetics of primary hepatocytes isolated from CREBH-KO and WT mice by analyzing oxygen consumption rate (OCR), which reflects OXPHOS, using Seahorse technology. The basal OCR of CREBH-KO mouse primary hepatocytes was higher than that of WT mouse hepatocytes (Fig. 7 D and E). During the mitochondrial stress test, the addition of the protonophore uncoupler FCCP (carbonyl cyanide-4 trifluoromethoxy phenylhydrazone) stimulated the OCR to its maximal activity in both WT and CREBH-KO cells, but to a significantly higher extent in CREBH-KO cells (Fig. 7 D and F). Moreover, the ATP-linked OCR, but not spare respiratory capacity or proton leak OCR, of CREBH-KO hepatocytes was significantly increased, compared to that of WT hepatocytes (SI Appendix, Fig. S9). This is consistent with the increased COX activities detected in the CREBH-KO mouse livers (Fig. 2 D and E). Further, we measured glycolysis with CREBH-KO and WT mouse hepatocytes. Concomitant with the increased OCR, CREBH-KO primary hepatocytes exhibited a significantly decreased glycolysis rate, compared to WT primary hepatocytes (Fig. 7G). This is consistent with the concept that UPRmt, which is defective in CREBH-KO hepatocytes, suppresses OXPHOS but increases glycolysis to facilitate energy production (22). Collectively, these results suggested that CREBH plays an important regulatory role in mitochondrial energy metabolism by modulating mitochondrial OXPHOS and glycolysis via UPRmt in response to energy-demanding challenges. CREBH deficiency leads to dysregulation of mitochondrial functions, characterized by a defective UPRmt (Fig. 4), increased OXPHOS but decreased glycolysis (Fig. 7), as well as decreased mitochondrial membrane potential and citrate synthase activities (Fig. 2 A and B), upon energy demands. To further delineate the regulatory role of CREBH in energy metabolism in vivo, we characterized the metabolic phenotype caused by CREBH deficiency with CREBH-KO and WT mice through the Phenomaster Laboratory Animal Monitor System. CREBH-KO mice exhibited reduced fat oxidation but increased glucose oxidation compared to the WT control mice (SI Appendix, Fig. S10 A–D). The increased glucose oxidation is in agreement with the increased mitochondrial OXPHOS and defective UPRmt in CREBH-KO mouse livers (Fig. 7A, and D–F). Interestingly, the level of energy expenditure of CREBH-KO mice was comparable to that of WT control mice (SI Appendix, Fig. S10 E and F). Consistently, the daily locomotor activities and food intake of CREBH-KO mice were comparable to that of WT mice (SI Appendix, Fig. S10 G–I), although the body weight of CREBH-KO mice was lighter (SI Appendix, Fig. S10J). Together, these results suggest that CREBH-KO mice have a change in fuel selection between fat and glucose, with decreases lipid utilization but increased reliance on carbohydrates for energy production. Discussion In this study, we revealed that the ER-tethered stress sensor CREBH functions as a major regulator of mitochondrial homeostasis and energy metabolism by activating hepatic UPRmt in response to energy demands (Fig. 7H). The identified CREBH-ATF4/5-UPRmt regulatory axis modulates mitochondrial OXPHOS and glycolysis and thus rebalances the metabolic landscape and energy production under energy-demanding challenges. Our major findings include: 1) CREBH is required to maintain mitochondrial homeostasis and function upon energy demands; 2) CREBH is enriched in and required for fasting-induced MAM expansion in the liver; 3) CREBH deficiency leads to accumulation of unfolded proteins in the mitochondria; 4) CREBH functions as a transcriptional regulator of UPRmt in the liver in response to fasting or under the circadian clock; 5) CREBH activates expression of ATF5 and ATF4, the master transcriptional regulators of UPRmt, in an UPRER-independent manner; 6) PPARα regulates expression of hepatic ATF5 in a CREBH-dependent manner; 7) CREBH regulates hepatic mitochondrial energy metabolism via the UPRmt-OXPHOS regulatory axis; and 8) CREBH deficiency decreases lipid utilization but increases reliance on carbohydrates for energy production. These findings not only defined a master regulator of UPRmt, but also revealed a crucial link between ER and mitochondria in maintaining mitochondria homeostasis and function under energy demands. The paradigm of regulation of mitochondria by an integrated ER-mitochondria stress signaling pathway, where CREBH, MAM, UPRmt, and OXPHOS form a regulatory network, has important implications in the understanding and treatment of metabolic or mitochondrial diseases. Increasing evidence suggests that stressed mitochondria interact with the ER, where organelle stress responses are integrated to modulate cell metabolism in response to energy fluctuations (11, 14). Mitochondria-ER contacts, known as MAMs, are composed of ER and mitochondrion membranes, housing many molecules that are situated at contact sites, and facilitate signal exchange between two compartments (28, 38, 39). As revealed by this study, CREBH protein is significantly enriched in MAMs of mouse livers in response to fasting (Fig. 3). CREBH is required for the expansion of fasting-induced MAM, as CREBH deficiency diminished the presence of stress-induced MAM components, including the primary UPRER transducers IRE1α and PERK as well as the mitochondrial fusion protein MFN2 and MAM-localized stress chaperone GRP75 (Fig. 3 B and C). MAM is a stress-inducible, dynamic interorganelle compartment, of which the components vary in a context-dependent manner (38, 39). The defect in fasting-induced MAMs, which is concomitant with the defective UPRmt, in CREBH-KO mouse livers suggests that CREBH functions as a major regulator of mitochondrial homeostasis and metabolism by mediating stress signaling pathways originating from the ER-mitochondria contact sites. CREBH acts as an indispensable regulator of both MAM and UPRmt triggered by fasting and therefore represents a critical node for the functional connection between MAM and UPRmt upon energy demands. Indeed, multiple CREBH-regulated UPRmt components, including PMPCB, LonP1, and TIMM17A, are involved in MAM formation or as a part of MAM components. Our study characterized a physiological UPRmt induced by energy demands associated with nutrient starvation or the circadian clock. The regulation of hepatic UPRmt by CREBH is likely limited to physiological stress conditions (Fig. 4), as CREBH failed to regulate UPRmt in response to chemical compounds that cause acute mitochondrial stress (SI Appendix, Fig. S11). Additionally, CREBH coordinates with PPARα to regulate UPRmt in mouse livers in response to fasting (Fig. 6). CREBH regulates and interacts with PPARγ coactivator 1α (PGC-1α), a transcriptional activator known to regulate mitochondrial homeostasis and metabolism, upon the energy-demanding challenge (1, 40) (SI Appendix, Fig. S12). It would be interesting to elucidate whether CREBH, PPARα, and PGC-1α regulate each other and function in synergy to regulate hepatic UPRmt and mitochondrial energy metabolism in the future. It has been suggested that UPRmt and its crosstalk with UPRER is regulated mainly through the PERK signaling pathway (11, 17, 41). In response to ER stress, the UPRER transducer PERK is activated to phosphorylate the translation initiation factor eIF2α, leading to a general protein translation attenuation but selectively increasing the translation of specific stress proteins (19). Particularly, ATF5, the master regulator of UPRmt, ATF4, and CHOP are induced via the PERK-eIF2α-UPRER pathway to activate UPRmt (11, 14, 17). As ATF4 and CHOP regulate expression of mitochondrial chaperone proteins and proteases by inducing ATF5, it has been proposed that there exists an intimate connection between ATF4, CHOP, and ATF5 and that PERK acts as a signal-transducing node connecting UPRmt and UPRER via the regulatory axis ATF4-CHOP-ATF5. Unlike the previously established UPRER-UPRmt regulatory model, our study revealed an alternative UPRmt signaling pathway mediated by a CREBH-ATF5/ATF4 regulatory axis in response to energy demands. CREBH directly binds to and activates transcription of the gene promoters of ATF5 and ATF4, but not CHOP or ATF3, in the livers upon fasting in a PERK- or UPRER-independent manner (Figs. 5 and 6 and SI Appendix, Fig. S5). The CREBH-driven activation of ATF5 and ATF4 led to up-regulation of the major UPRmt components, suggesting a physiological UPRmt pathway mediated through the CREBH-ATF5/ATF4 regulatory axis. Interestingly, ATF4 is known to induce expression of CHOP, a stress-induced transcriptional regulator that acts pleiotropically as a proapoptotic or proadaptive factor during ER stress (19, 42). However, our studies suggest that CREBH activates expression of ATF5 and ATF4, but not CHOP, in response to the fasting challenge, in a manner independent of the PERK-mediated UPRER pathway. CHOP is unlikely to be a target of ATF4 in the CREBH-driven UPRmt regulatory cascade under the mitochondrial stress triggered by energy demands. How ATF4 selectively induces expression of CHOP in a context-dependent manner is an interesting question to be explored in the future. Our study revealed that CREBH regulates mitochondrial energy metabolism through regulating UPRmt. The regulatory role of UPRmt in mitochondrial metabolism is mainly reflected in the synchronous regulation of OXPHOS and glycolysis. During mitochondrial stress, UPRmt limits transcription of OXPHOS components to reduce the load on the mitochondrial proteostasis machinery but increases glycolysis to promote an alternative means of ATP production (21, 43). Indeed, the defective UPRmt in CREBH-KO mouse livers was coincident with the increased expression of OXPHOS components, up-regulation of OCR, decreased expression of glycolysis genes, and down-regulation of glycolysis in CREBH-KO mouse livers or primary hepatocytes (Fig. 7). Our data suggested that CREBH plays an important role in modulating OXPHOS and glycolysis via activating UPRmt during mitochondrial stress and thus rebalances the cellular metabolic landscape to facilitate mitochondrial recovery. CREBH-KO mouse livers displayed elevation of cellular ROS signals under the fasting challenge (Fig. 2 F and G). Although UPRmt was suggested to suppress mitochondrial ROS production during stress (14), the decreased hepatic mitochondrial membrane potential of CREBH-KO mice counteracted the possibility of increased hepatic mitochondrial ROS by CREBH deficiency (Fig. 2B). Whether CREBH deficiency or dysregulated mitochondria/MAM elevates cytosolic ROS production remains to be elucidated. Nevertheless, increased hepatic ROS production in CREBH-KO mice under energy demands represents a major risk factor for cardiovascular and metabolic disorders. Human genome-wide association studies (GWAS) analysis showed that human CREBH gene variants are linked to the incidences of hypertriglyceridemia, cardiovascular disease (CAD), and type-2 diabetes (T2D) (SI Appendix, Table S1). CREBH common gene variants were associated with the major parameters of cardiovascular and metabolic disorders, including elevated body mass index (BMI), visceral fat volume, body weight, as well as dysregulated levels of triglycerides and cholesterols. These phenotypes are consistent with impaired UPRmt and dysregulated mitochondrial function and homeostasis in the livers with CREBH deficiency. The key role of CREBH in regulating UPRmt and subsequent mitochondrial homeostasis and function could partially account for the metabolic phenotypes caused by CREBH deficiency. Not only engaging the pathogenesis of metabolic disorders, CREBH may also be involved in mitochondrial diseases. Therefore, the roles of CREBH in safeguarding against complex diseases merit further investigation. Materials and Methods MAM, ER, and Mitochondrial Protein Fractionation. Isolation of highly purified MAM, ER, and mitochondrial protein fractions from mouse liver tissues was described previously (44). Briefly, mouse liver tissue was homogenized, followed by centrifugating the homogenate at 6,300×g to collect nuclei-free crude ER, mitochondria, MAM, and cytosol fractions, respectively. The crude mitochondria/MAM fraction was centrifuged at 95,000 g in an ultracentrifuge to separate a mixture of MAM and fragmented mitochondria (upper band) and mitochondria (lower band) (SI Appendix, Fig. S13). The MAM-containing supernatants were centrifuged at 100,000×g to collect purified MAMs at the bottom. The lower mitochondria band was centrifuged at 10,000×g to collect purified mitochondria. The crude ER-containing cytosolic supernatant was centrifugated at 20,000×g to separate the ER fraction in the supernatant, followed by centrifuging the supernatant at 100,000×g to collect the purified ER fraction at the bottom. Measurements of Mitochondrial Citrate Synthase (CS) Activities, Membrane Potential, ATP Concentrations, and COX Activities. Purified, intact mitochondria were isolated from the livers of CREBH-KO and WT mice under the feeding or fasting condition and then subjected to measurements of mitochondrial citrate synthase (CS) activities, mitochondrial ATP concentrations, as well as mitochondrial cytochrome c oxidase complex IV (COX) and complex I enzyme activities using the kits commercially available. Mitochondrial ATP levels were quantified by inactivation of ATPases in purified mitochondrial fractions, followed by measuring ATP levels based on the production of light caused by the reaction of ATP with added firefly’s luciferase and luciferin. Mitochondrial membrane potential was evaluated by fluorescence intensity using JC-1, a lipophilic dye (MedChemExpress Inc), and indicated by ratios of aggregate vs monomer JC-1. For all the measurements, mitochondrial protein concentrations were quantified for normalization of mitochondrial CS activities, ATP concentrations, membrane potentials, COX activities, and complex I activities. For a full description of Materials and Methods used in this work, SI Appendix, Supplementary Information. Supplementary Material Appendix 01 (PDF)